Vincent Sanders
c9ee49baa8
cope with backing store writeout making no progress
2014-12-01 11:32:33 +00:00
Vincent Sanders
ba5ade87b2
ensure the backing store write reports at least a minimal amount of time
2014-12-01 09:18:31 +00:00
Vincent Sanders
ca74e0060d
release gadget allocation on error path (Coverity 1109875)
2014-11-30 11:23:30 +00:00
Vincent Sanders
1c735140b6
fix resource leak on error (Coverity 1256245)
2014-11-30 11:21:34 +00:00
Chris Young
6a2461da8d
Enable the async write process to be disabled for testing purposes
2014-11-30 01:25:20 +00:00
Vincent Sanders
9bc07e8d8f
make coverity build script use updated environment variables
2014-11-30 00:37:35 +00:00
Vincent Sanders
61ab2777da
Improve writeout rescheduling if list of objects is completed quickly
2014-11-30 00:27:05 +00:00
Vincent Sanders
d08acbc5ae
correctly calculate writeout bandwidth and properly impose limits
2014-11-29 23:57:15 +00:00
Chris Young
70e777c30d
Add some comments about timer.device.
2014-11-29 19:56:54 +00:00
Vincent Sanders
6b2676d33f
Improve llcache writeout stratagy
2014-11-28 19:17:42 +00:00
Vincent Sanders
5f57a39130
revert unecessary librt linkage as it not comes from libnsfb pkg-config correctly
2014-11-27 16:04:16 +00:00
Vincent Sanders
18d1fce332
add rt library to link
2014-11-27 14:40:00 +00:00
Vincent Sanders
8d51dd4879
update the backing store to allow for invalidating entries which have an allocation
2014-11-27 13:31:43 +00:00
Vincent Sanders
79d5b1bf3b
stop using gettimeofday and use the monotonic time interface fron nsutils
2014-11-26 21:29:13 +00:00
Vincent Sanders
81417ae4a5
caclulate the total amount of time spent writing an object out
...
In order to calculate the writeout bandwidth we need to know how long
it took to write the data to peristant storage in addition to how much
was written.
2014-11-25 23:11:41 +00:00
Vincent Sanders
cd2e336449
Try using -pipe on beos/haiku as temporary file access may be slow
2014-11-24 13:18:52 +00:00
Michael Drake
b9a6dae5ab
Remove unecessary header include.
2014-11-24 09:43:59 +00:00
Vincent Sanders
5a78cd810f
correctly release source data from llcache
2014-11-23 18:21:56 +00:00
Vincent Sanders
01b2bc5a57
Improve doxygen comments and add todo for cache and backing store
2014-11-23 15:43:03 +00:00
Vincent Sanders
246a1fec05
remove constness from backing stores store method
2014-11-23 01:37:24 +00:00
Vincent Sanders
63c35da66e
annotate element struct with packing attributes to reduce entry size
2014-11-23 01:25:14 +00:00
Vincent Sanders
08c7ef6bed
Do not use anonymous union as it is not portable
...
unnamed unions are not portable to old compilers. When reloading data
ensure the heap allocation status is discarded.
2014-11-23 00:35:11 +00:00
Vincent Sanders
8b810ee4a1
change the persistant data store to owning the allocations
2014-11-22 23:56:43 +00:00
Chris Young
335ba082fd
Ensure we're not running events ahead of time.
2014-11-22 17:15:52 +00:00
Chris Young
4a395ad025
Process scheduled event changes before incoming timer events.
...
This stops things being run that have just been deleted. However, may cause the next event to run early because we don't check the time.
2014-11-22 17:10:07 +00:00
Chris Young
4589cab433
Re-enable asynchronous backing store writes
2014-11-22 16:52:27 +00:00
Chris Young
bcd7413f8e
Fix llcache ops table
2014-11-22 16:51:00 +00:00
Chris Young
0cd682644b
Revert "Revert "Write out to the backing store asynchronously.""
...
This reverts commit 75623179aa
.
2014-11-22 16:41:29 +00:00
Chris Young
71cdc991c7
fix warning
2014-11-22 16:39:31 +00:00
Chris Young
277ceb9fc2
Reduce stack to 2K
2014-11-22 16:38:26 +00:00
Chris Young
9e98d77531
Reply to the timer before running the scheduled task.
2014-11-22 16:37:49 +00:00
Chris Young
186e1f4ee3
New asynchronous scheduler
...
This ensures that if other processes other than the main NetSurf process try to create scheduled tasks, they are always run on the main process.
2014-11-22 16:30:43 +00:00
Chris Young
2de1553a00
New scheduler
2014-11-22 16:27:39 +00:00
Chris Young
f8d8d218d8
Initial steps for a scheduler process
2014-11-22 16:27:39 +00:00
Chris Young
fe56795205
Revert "Should be a while loop in case of multiple simultaneous event triggers."
...
This reverts commit 9dcf001a0b
.
For ehatever reason this is entirely broken, but I'm in the process of rewriting so just revert for now to get the working builds again.
2014-11-22 14:18:00 +00:00
Michael Drake
d9a63bb9e0
Fix brackets.
2014-11-22 00:06:01 +00:00
Michael Drake
1a1b6f8984
Shouldn't need browser_private.h now.
2014-11-22 00:02:28 +00:00
Michael Drake
0756b544c6
Use API to find if content is selectable, instead of guessing.
2014-11-22 00:02:28 +00:00
Michael Drake
ba9a049d7b
Simplify ami_history_open API and use correct call to get history object.
2014-11-22 00:02:28 +00:00
Michael Drake
0c7dc95dda
Use API to get URL.
2014-11-22 00:02:28 +00:00
Michael Drake
c9683119c3
Use browser_window_get_content where we need to.
2014-11-22 00:02:27 +00:00
Chris Young
9dcf001a0b
Should be a while loop in case of multiple simultaneous event triggers.
2014-11-21 23:47:26 +00:00
Michael Drake
60c33e0410
Should be free of reliance on browser window internals here now.
2014-11-21 14:39:12 +00:00
Michael Drake
781fe8f330
Use browser_window_get_content() to get the current content.
...
Although I dislike this API, and it may go at some point.
2014-11-21 14:38:17 +00:00
Michael Drake
90f89f98e3
Use API for getting browser window history.
2014-11-21 14:37:17 +00:00
Michael Drake
b4e044f6c9
Use browser_window_get_url() instead of getting URL from current content.
2014-11-21 14:30:09 +00:00
Michael Drake
3b686eaccc
Use browser_window_can_select(), instead of assuming based on content type.
2014-11-21 14:26:29 +00:00
Vincent Sanders
9fb0eaa1cd
fix missing semicolon
2014-11-21 13:14:20 +00:00
Michael Drake
19af487fa1
Shouldn't need browser_private.h now.
2014-11-21 12:56:19 +00:00
Michael Drake
47f6e29b4c
Use browser_window_stop_available()
...
Looking at bw->loading_content was bad and wrong in the cases
of frames, and object fetches.
2014-11-21 12:53:55 +00:00